Using Gene Analytics to Identify Patients at Risk for Treatment Toxicity

A Conversation With Stephen T. Sonis, DMD, DMSc

Get Permission

Stephen T. Sonis, DMD, DMSc

Jamie H. Von Roenn, MD

To throw the biggest net possible, we’ve studied the most common form of genetic variations—single nucleotide polymorphisms—and identified specific groups of them that predict regimen-related side effects.

—Stephen T. Sonis, DMD, DMSc

Genomic applications are now an accepted part of oncologic science and practice. Differences in gene expression have been used to understand and predict tumor behaviors and response to treatment. And now it seems likely that genomics may also play a pivotal role in guiding treatment preferences by identifying a patient’s risk for treatment toxicities. Understanding the fundamental pathobiologic processes triggered by cancer drugs or radiation therapy, identifying the genetic factors that predispose patients to treatment toxicities, and developing effective interventions to counter those toxicities are the focus of the laboratory and clinical research of Stephen T. ­Sonis, DMD, DMSc.

Although Dr. Sonis’ earlier career centered on the biologic pathways that are associated with the generation of severe cancer regimen–related oral mucositis, his current research has expanded to include a broad range of treatment-induced side effects, including fatigue, fibrosis, neuropathy, nausea and vomiting, diarrhea, and cognitive dysfunction. Dr. Sonis wrote about the application of genomics in the supportive care of patients with cancer in the 2015 ASCO Educational Book.1

The ASCO Post talked with Dr. Sonis, Senior Surgeon, Brigham and Women’s Hospital and Dana-Farber Cancer Institute; Professor, Department of Oral Medicine, Infection, and Immunity, Harvard School of Dental Medicine; Chief Scientific Officer of Biomodels, LLC; and Cofounder and Chief Scientific Advisor of Inform Genomics, a privately held company, about the role genomics plays in the development of cancer treatment toxicities and how to identify patients who are at risk.

Biologic Toxicity Triggers

What is your research showing regarding the role genomics plays in determining whether a patient will develop side effects from cancer treatment?

When we talk about using genomics to identify a patient’s risk of regimen-related toxicity, the low-hanging fruit has typically been associated with genes that control the metabolism of cancer drugs—pharmacokinetics.

For example, if a patient who is being treated with fluorouracil (5-FU) has a genetically controlled deficiency in one enzyme (dihydropyrimidine dehydrogenase [DPD]) that metabolizes the drug, he or she will accumulate levels of the drug that will increase the likelihood of the development of associated toxicity. In fact, a test for the gene associated with the enzyme, TheraGuide 5-FU, was approved some time ago.

Enzyme deficiencies, however, are relatively rare compared with the overall number of patients who develop regimen-related toxicities. The more common risk drivers of genomic toxicity, whether from chemotherapy, radiation therapy, or targeted therapy, are associated with the genes controlling the biologic actions of anticancer treatment. These genes are usually associated with specific pathways, which may act alone or synergistically to cause a cascade of disruptive biologic signals that target key cells in normal tissue to produce damage.

We now know that many tissue-based chemotherapy-associated toxicities, such as oral mucositis, esophagitis, proctitis, pneumonitis, radiation-induced dermatitis, and fibrosis, share common genetically controlled underlying biologic etiologies. Interestingly, other toxicities, such as cognitive dysfunction and fatigue, are impacted by many of the same biologic mediators. Understanding these biologic toxicity triggers initiated by chemotherapy or radiation therapy allows us to look for the patient-specific differences in genes that are associated with risk.

Single Nucleotide Polymorphisms

How do you identify patients who are at greater risk for treatment side effects?

Our goal is to prospectively identify the genetic and nongenetic elements that define an individual’s risk for side effects. To distinguish the genomic factors, our approach has focused on the premise that risk is best defined by groups of cooperating genes, rather than a single “star.”

To throw the biggest net possible, we’ve studied the most common form of genetic variations—single nucleotide polymorphisms—and identified specific groups of them that predict regimen-related side effects. Although there are “only” 25,000 genes, there are about 10 million single nucleotide polymorphisms, which means that we have an enormous big-data challenge to evaluate the associations between these gene variants and the patient’s reported toxicities.

Since single nucleotide polymorphisms are a component of DNA, they are easily attainable via a saliva sample that can be used to extract DNA. To effectively translate these concepts to patients, we cofounded Inform Genomics in July 2010. The company is currently developing precision medicine products designed to accurately create a predictive toxicity risk profile for patients about to undergo chemotherapy for the treatment of solid tumors or for individuals about to receive high-dose chemotherapy conditioning regimens prior to hematopoietic stem cell transplantation.

Results from two preliminary studies2,3 have confirmed our ability to identify clusters of single nucleotide polymorphisms that are associated with an increased likelihood of a range of common treatment side effects in both treatment populations.

Clinical Use of Gene Analytics

When will this type of gene analytics be ready for clinical use?

I’m hoping within 2 to 3 years.


How do you envision genomic analytics will actually be used in the clinic?

Identification of toxicity risks prior to the initiation of therapy will provide the oncologist and patient with critical information that can help guide treatment choices. For example, multiple drug regimens are available for almost all tumor types, each with unique toxicity profiles. For example, neuropathy might be common with one, whereas gastrointestinal toxicities may dominate another. For a patient who uses a computer keyboard, the first option might be intolerable, whereas in a frail individual, the consequences of gastrointestinal toxicities might preclude the second option.

Although it may not be possible to completely prevent the onset of toxicities, knowledge of risk will provide oncologists and patients with actionable information for individual risk-based consent and cost-effective preemptive supportive care measures, patient education, and more frequent surveillance for patients at high risk for treatment toxicities. ■

Disclosure: Dr. Sonis is Chief Scientific Officer of Biomodels, LLC, and Cofounder and Chief Scientific Advisor of Inform Genomics. He also is on the medical advisory boards of several biotechnology and pharmaceutical companies.


1. Sonis ST: Genomics, personalized medicine, and supportive cancer care. 2015 ASCO Educational Book. Available at Accessed July 8, 2015.

2. Sonis S, Antin J, Tedaldi M, et al: SNP-based Bayesian networks can predict oral mucositis risk in autologous stem cell transplant recipients. Oral Dis 19:721-727, 2013.

3. Sonis ST, Schwartzberg LS, Walker MS, et al: Predicting risk of chemotherapy-induced side effects in patients with colon cancer with single-nucleotide polymorphism Bayesian networks. 2013 Gastrointestinal Cancers Symposium. Abstract 344.

Addressing the evolving needs of cancer survivors at various stages of their illness and care, Palliative Care in Oncology is guest edited by Jamie H. Von Roenn, MD. Dr. Von Roenn is ASCO’s Senior Director of Education, Science, and Professional Development Department.